{"id":"W7135988917","doi":"","title":"Recall Aspects of Transformers for Text Ranking","year":2022,"lang":"en","type":"article","venue":"UvA-DARE (University of Amsterdam)","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Canadian Institute of Steel Construction","keywords":"Pooling; Recall; Transformer; Deep learning; Ranking (information retrieval); Precision and recall; Artificial neural network; Deep neural networks","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003179601,0.00007523559,0.000174527,0.0002095931,0.0003411743,0.00001401654,0.0007028273,0.00003129975,0.0002589377],"category_scores_gemma":[0.000008319886,0.00009156445,0.000161506,0.0004215606,0.00006841031,0.0005452725,0.0002270703,0.0001167056,0.000005611862],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007219225,"about_ca_system_score_gemma":0.0001257629,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008593348,"about_ca_topic_score_gemma":0.00001994106,"domain_scores_codex":[0.9990634,0.00004396176,0.0001419662,0.0001562878,0.0004000457,0.0001943739],"domain_scores_gemma":[0.9994183,0.00006500887,0.0001330693,0.0001741629,0.0001472735,0.00006217448],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0009199857,0.0005151961,0.0009712774,0.0006290231,0.0001716777,0.0000653992,0.06342597,0.0007184908,0.01156912,0.1375843,0.004498667,0.7789309],"study_design_scores_gemma":[0.02906619,0.01066883,0.05439816,0.0003760611,0.0003779768,0.0001784276,0.08653689,0.2147018,0.03715571,0.01280522,0.550608,0.003126684],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3123563,0.00003472136,0.6644359,0.001885602,0.0003503186,0.000694547,0.0001525283,0.00009349881,0.01999667],"genre_scores_gemma":[0.9924665,0.000005419367,0.006595598,0.00007681263,0.000006618126,8.222902e-7,0.0000153839,0.000003944407,0.0008288889],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7758042,"threshold_uncertainty_score":0.373389,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01779624396452778,"score_gpt":0.2138090191995263,"score_spread":0.1960127752349985,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}